Codex Desktop analytics queue saturation emits thousands of warnings and drops events
Nobody has claimed this yet.
- Dominant language
- Rust
- Stars
- 125k
- Forks
- 19.5k
- PR merge metrics
- PR metrics pending
Description
Summary
Codex Desktop can saturate its internal analytics queue during a tool-heavy, multi-task session and then repeatedly drop analytics events.
A read-only aggregate of the local structured log recorded 4,050 WARN entries from codex_analytics::client containing:
dropping analytics events: queue is full
The warnings accumulated in roughly one hour.
Environment
- Codex Desktop:
26.803.61601 - Bundled CLI:
0.147.0-alpha.6.5 - Platform: macOS arm64
- Workload: long-running desktop session with multiple active tasks and frequent tool calls
Steps to reproduce
- Keep Codex Desktop open with multiple active tasks.
- Run tool-heavy turns for an extended period.
- Inspect the structured Codex log for the
codex_analytics::clienttarget. - Count messages containing
dropping analytics events: queue is full.
Actual behavior
The queue remains saturated long enough to emit thousands of warnings and explicitly drops analytics events. The warning volume also adds substantial log noise.
Expected behavior
The analytics producer should apply bounded backpressure or the consumer should drain fast enough that routine multi-task use does not continuously drop events. Repeated identical warnings should also be rate-limited.
Privacy
This report contains only aggregate counts, versions, and the stable warning text. Local paths, task and call identifiers, account details, configuration values, and unrelated logs are omitted.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start at the codex_analytics::client target and reproduce the multi-task workload while inspecting the structured log for queue saturation. Trace the analytics producer and consumer to determine where backpressure and warning repetition occur. Done means routine use no longer continuously drops events and identical queue-full warnings are rate-limited.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust
- Domain
- observability
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100